Skip to content
US GenAI Updates
US GenAI Updates
Historical Updates
Share
Explore

3rd July, 2025

AutoAI Updates

Veo3 + Playground Integration:
Preview videos and generation history are deployed inside Playground
Adoption remains at ~40-45 ACDs using it on a daily basis.
Core reason:
LoRA Execution and upgrades:
Training dataset upgrade: New 2K resolution, multi‑angle dataset stabilizes character appearance and facial range.
Feedback from CDs is due this week; if results hold up, character consistency issue will be resolved.
Slugline Execution:
Writers have been trained on cleaner sluglines; blank images now appear only when scenes are extremely NSFW.
Training Guide:
Slugline guide
⁠
⁠
Background Consistency:
Long assets (> 30 min) sometimes drop sluglines and shift colors in the background.
Tech fix in progress; ETA – Fri, 4 Jul
ChatGPT-4o Integration: MVP has been launched
V1 is live; currently facing issues with blank images due to content policy violation (not limited to NSFW).
Costume consistency will be incorporated with ChatGPT-4o V2 – ETA pending.
Experiments:
Experiment Sheet Link
⁠
⁠
⁠

AutoAI Issues to Resolve in the Next 30 Days

Blank images in ChatGPT-4o
Costume inconsistency – awaiting ChatGPT-4o V2 deployment. Will pick this post that.
Background errors on long assets (above 30 mins)
Low visual dynamism & repetitive close‑ups
Limited expression range in some faces
⁠

New Tools Being Tested

Seedance 1.0: Better than Veo 3 in prompt adherence and camera control
Strong camera control and multi‑shot support; no audio output and lower cost than Veo 3.
Next: run one canvas test with character prompts to confirm day‑to‑day viability.
Samples:
Seedance output
⁠
⁠
⁠

Training Sessions

29 Jun – Parth  & Ashok: Custom GPT Models for Prompt Creation for Fantasy Scenes (
link
⁠
)
Slugline Writing Refresher – 24 Jun (internal) -
Async Loom
⁠
⁠

Want to print your doc?
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.